Addressing Substance use Disorder in Primary Care: The Role, Integration, and Impact of Recovery Coaches

Background: Only 10% of people with substance use disorder (SUD) receive treatment, partially due to inadequate access to specialty SUD care and limited management within primary care. “Recovery coaches” (RCs), peers sharing the lived experience of addiction and recovery, are increasingly being integrated into primary care to help reach and treat people experiencing SUD, yet little is known about how their role should be defined or about their clinical integration and impact. Methods: Semistructured interviews with RCs (n = 5) and their patients (n = 16) were used to explore patient and RC perspectives on the RC role. Maximum variation sampling was employed to select patients who displayed diversity across gender, RC, housing status, and number of contacts with an RC. Patients were sampled until no new concepts emerged from additional interviews, and a semistructured interview guide was used for data collection. To analyze interview transcripts, the constant comparative method was used to develop and assign inductively developed codes. Two coders separately coded all transcripts and reconciled code assignments. Results: Four core RC activities were identified: system navigation, supporting behavior change, harm reduction, and relationship building. Across these activities, benefits of the RC role emerged, including accessibility, shared experiences, motivation of behavior change, and links to social services. Challenges of the RC model were also evident: patient discomfort with asking for help, lack of clarity in RC role, and tension within the care team. Conclusions: These findings shed light on RCs in primary care. Many patients and coaches perceived that RCs play a valuable role within primary care, providing both tangible system navigation and intangible, social support that promote recovery and might not otherwise be available. Enhanced communication between RCs and health center leadership in defining the RC role may help resolve ambiguity and related tensions between RCs and care team members.


Introduction
In the United States, only 10% of the 22 million people with a substance use disorder (SUD) receive treatment. 1 SUDs drive up costs of care: nationwide, more than 1 in 5 patients hospitalized on general medicine wards have SUD, 2 and people with SUDs are 1.5 times more likely to be readmitted to the hospital. 3,4 Faced with this challenge, primary care providers (PCPs) can provide effective treatment with medications for addiction treatment, but few PCPs offer this treatment, citing lack of psychosocial and logistic supports as barriers. 5,6 To address these barriers, one addition to SUD management has been the emergence of "recovery coaches" (RCs). RCs are typically individuals in recovery from SUD who are employed to assist and provide guidance to patients in various stages of recovery. 6,7 Also known as "peer support specialists" and "certified peer specialists," 8 they have played a variety of roles, 9 including supporting people in recovery within community organizations; serving as one-on-one or small group supporters for private addiction patients 10 ; and delivering brief interventions in emergency departments. 7 Peers have long been part of SUD recovery, yet there is a key distinction between "peer support" or "mutual help" and "peer providers." In the case of "peer support" and "mutual help," all participants are giving and getting support in their recovery. One example of this is 12-step organizations, where peer mentoring ("sponsorship") is a key tenant. 11 "Peer providers," such as RCs, are part of professional SUD services and often combine lived experience with formal training. 8 Like all health care providers, the support is primarily unidirectional. The provider supports the patient in her recovery but does not get support from the patient. 10 Peer interventions have been effectively used for management of other chronic diseases, [12][13][14] but the formalized peer role is newer to SUD services.
The existing literature on recovery coaches is sparse and difficult to interpret because the studies used widely disparate definitions of the RC role. 9 Because there has been little research on RCs in primary care settings, we looked across settings to understand what is known about the outcomes of RCs. A recent systematic review examining outcomes of "RC" interventions found relatively few studies and inconsistent definitions of RCs across studies, including inconsistent emphasis on prior personal experience with substance use. Despite the variation in settings, populations, and RC responsibilities in prior research, overall, peer provider interventions tended to have positive effects on patient recovery and other outcomes, 15 including legal charges 16 and use of health services. 17,18 RCs embedded within primary care may have the greatest impact on the addiction treatment gap, as they have the potential to make SUD care accessible outside of a specialist setting. However, without much prior research on how to integrate RCs into primary care, the RC role in this setting remains poorly defined and potentially wide in scope. 9 Greater knowledge regarding patients' and RCs' experiences could inform the design and implementation of peer provider models in primary care settings.
Our institution, a large academic hospital system, recently hired RCs and embedded them within primary care. If a primary care physician made a clinical diagnosis of SUD (any substance), the physician referred that patient to a RC by introducing the RC during a clinic visit or asking the RC to reach out to the patient. The coaches had on-site clinical supervision from a social worker or physician and weekly supervision sessions with a physician and psychologist to discuss coach well-being and shared clinical or systems challenges, such as difficulty accessing resources for a patient.
This study employed qualitative interviews with RCs and their patients to understand and define the RC role within a primary care team. Qualitative methods facilitate in-depth exploration of the complexity of the RC role and relationships among RCs, patients, and other team members. Such data can be used to develop a consensus definition of RCs as a new member of the health care workforce and inform how RCs are trained, utilized, and regulated.

Methods
Study reporting is guided by the consolidated criteria for reporting qualitative research (COREQ). 19

RC program and setting
Interviews were conducted within recently developed recovery coach initiative at Massachusetts General Hospital (MGH) and its community health centers. MGH is an ideal site for an exploratory study of recovery coaching practices given that the MGH RC program is large and among the first to integrate RCs into primary care teams across a hospital system. At the time of the study, the hospital system employed 5 RCs, 1 at each of 3 community health centers, 1 at a clinic for the homeless population, and 1 embedded with the inpatient addiction consult team but who often worked with previously hospitalized patients in an outpatient setting. RCs were required to have at least 2 years of sobriety (no use of substance(s) of prior misuse or dependence) by self-report. Some coaches were taking medications for addiction treatment. All coaches took a standardized, 5-day course in recovery coaching led by the state's health department. The training included self-care, an overview of recovery modalities, and basics of motivational interviewing. 20 Coaches worked full time and were salaried ($35,000-$50,000 per year depending on degrees and experience) with benefits.

Study design and sample
This is a qualitative, descriptive study. 21 We conducted individual interviews with RCs (n D 5) and their patients (n D 16). All of the coaches who had been employed for at least 4 months at the time of data collection were interviewed. We used maximum variation sampling to select patients, 22 identifying patients who displayed diversity across gender, housing status, the specific RC they worked with, and number of RC contacts. Maximum variation sampling is appropriate to select patient participants for this study because it helped us elicit a breadth of perspectives, rather than a representative sample. 22 As we aimed to explore the way RCs fit into the larger care team, we recruited only patients who were connected to a primary care provider at an MGH site at the time of the study. We did not screen patients for SUD. We relied on the clinical judgment of the physicians who diagnosed them with an SUD before referring them to an RC. Although there is considerable evidence that primary care providers underdiagnose SUD, 23,24 it is unlikely that patients without SUD would be referred to an RC.
No patients declined to participate, although 2 interested patients were unable to be reached for an interview. Patients were given a $15 gift card and coaches a $40 gift card for participating. All coaches were interviewed. Patients were interviewed to theoretical saturation, the point at which no additional concepts emerged from subsequent interviews. 25,26 Informed consent was obtained from all participants. The Partners Human Research Committee approved this study (Protocol 2015P000072/MGH).

Data collection
Prior to the interview, the interviewer told RCs that the goal of the study was to better understand the RC program and explained to patients that the goal was to understand SUD care. The interviewer clarified that the study was being conducted for research and program evaluation and that she had no role in employment (RCs) or care (patients). Only the interviewer and participant were present during the interview.
We used semistructured guides with follow-up prompts to conduct the interviews (Appendix A). The interview guides and prompts encouraged respondents to share specific experiences, which may provide greater specificity and accuracy than general statements. A single interviewer (H.J.) who had no prior clinical relationship with any of the participants and was trained in qualitative interviewing conducted and recorded all interviews. 27 Each participant was interviewed once, and interviews lasted between 30 and 90 minutes. Interviews were professionally transcribed verbatim.

Data analysis
We used the constant comparative method for analysis of interview data. 25,28 Three researchers (H.J., D.O., S.W.) separately assigned codes to 4 transcripts. Inductive coding was used 28 ; each time a researcher came across a new idea in the transcript, she assigned it a label or "code," then assigned the same code to other quotes about the same idea. By the end of the 4 transcripts, no new codes were emerging, suggesting that it was not necessary to use additional transcripts to develop the preliminary code list. The researchers then reviewed the 4 transcripts together, blending their code lists into a single, comprehensive list.
Two researchers (H.J., D.O.) separately coded each transcript, applying the code list, but alerted for new ideas that should be added to make the code list comprehensive. One coder was a primary care resident (male) and the other a medical student (female), both with training in qualitative methods and experience working with patients with SUD. To ensure consistency within codes, coders referred to previously coded text. 25 During coding, an additional idea emerged that was not captured in the initial code list and was added. Prior transcripts were reviewed for instances of the idea. 26 Coded transcripts were entered into ATLAS.ti (version 1.0.38; Scientific Software Development, Berlin, Germany) to organize and retrieve data and review quotes within codes. Participants did not review transcripts or findings. The research team used the coded data to (1) combine codes into a smaller set of broader ideas; (2) identify codes that presented an idea that would not be sufficiently captured if combined with other codes; and (3) identify themes, cross-cutting ideas seen in all or many codes. 29 The researchers frequently returned to the original data to ensure that they represented the ideas of the participants. 30 A preset theoretical framework did not guide data collection and analysis. Rather, ideas that emerged from the study were used to guide theory in the discussion. 28

Results
Patient characteristics are displayed in Table 1. Because of the small number of coaches, we have not displayed coach characteristics to protect confidentiality.
Interviews with both coaches and patients revealed 4 types of coach activities, each of which were separate codes during data analysis: (1) system navigation: helping the patient access treatment, assisting with applications for social services, and accompanying the patient to court or medical appointments; (2) supporting behavior change: eliciting and sustaining discussions with patients about changing behaviors in multiple wellness domains, including substance use and diet; (3) harm reduction: providing patients with clean needles and naloxone kits and helping homeless patients get clothing and food if they did not want to be housed; and (4) relationship building: spending time with the patient without a specific agenda, often on the street or over a meal, coffee, or a cigarette. Quotes illustrating examples of each of these activities are displayed in Table 2.
Drawing on broad combinations of codes and themes emerging across codes, we discuss the strengths and challenges of the RC role that may influence patient outcomes and care team dynamics ( Table 3).

Accessibility
Many patients emphasized that their coach was easy to reach via phone or text message. Coaches responded more rapidly than a physician and could meet with patients at coffee shops or parks closer to the patient's home. Some patients appreciated that the coach was often at their primary care clinic and available to provide support during a challenging appointment.
The coaches acknowledged the need to balance being accessible to patients and "set boundaries." Although some patients perceived that coaches were always available, all but one coach said that they tried not to take patient calls at night or on weekends. Coaches also described efforts to proactively make contact with patients. For instance, to reach patients without access to a phone or who were hesitant to contact the coach, coaches would walk around the neighborhood or go to areas where homeless patients spent time.

Shared experiences
Many patients reported valuing shared experiences with the coaches. A number of patients explained that it was clear from their first meeting that the coach had "street cred," even if the coach revealed little about his or her own substance use history. They felt that their physician, who lacked a shared language or experiences of addiction, was more "exalted" and could not provide "practical advice." When coaches provided guidance, such as identifying situations in which the patient might feel tempted to use drugs, some patients took it more seriously than advice they got from their doctor because it was grounded in personal experience rather than "book learning."  Most coaches said that they did not freely share their personal experiences with patients. They would generally tell patients that they had struggled with addiction, revealing specific details only to advance patient well-being. For instance, when she thought a patient was about to use, a coach shared a story about her experience in a similar situation but did so only because she "had to pull something out to keep [the patient] engaged." The patient did not relapse. Many coaches perceived that sharing made the relationship about their own recovery and less about the patient.
Motivating behavior change Some patients felt that positive behavior was easier with their coach's support. Patients explained that they felt more at ease going to court and were more likely to attend hearings if their coach accompanied them. Furthermore, many patients praised their coaches for confronting them when they saw them engaging in risky behavior. They appreciated that their coach, unlike their physician, knew when they were being risky: "if you are screwing up, [they] will tell you." Echoing the remarks of their patients, when they shared specific experiences or stories some coaches discussed the importance of directly telling patients not to use and to be truthful. Conversely, if RCs spoke in general terms about their role, they almost uniformly said that they needed to allow patients to drive change: "it's not about lecturing, it's about support."

Links to local social services
Many patients spoke about how much they appreciated that their coach helped them connect with services, including housing, legal services, and addiction treatment. Some patients contrasted the role of the coach, who understood how and had time to navigate services, with that of the physician, who did not have time or would not know how. One patient explained that he was more motivated to stay in treatment because he knew his coach worked hard and used personal connections ("put her neck on the line") to get him into a treatment center.
Many coaches explained that helping patients access services was one of the most important aspects of their jobs, allowing patients to overcome specific "barriers" to functioning. All coaches spoke about the importance of being knowledgeable about local treatment services and having relationships with people who work there. Many of the coaches had worked in treatment facilities and remained connected to program staff. These personal connections enabled them to place patients in facilities that may not otherwise have accepted a patient. Coaches also used visits to social service agencies or court as opportunities to take the patient for a meal or to have coffee and talk.

Challenges of the RC role
Patient discomfort with asking for help Many of the patients who did not connect well with their coach explained that they felt uncomfortable reaching out for help, which they blamed on themselves. One patient, struggling with her search for housing, feared the coach would be discouraged by her lack of progress and was therefore hesitant to reach out for help. Others noted feeling uncomfortable with the coach because of differences between them, including the coach's age, gender, and substance use history.
In addressing these barriers, most coaches felt that they could not force patients to engage, but they shared examples of repeatedly calling, texting, or visiting patients: "your job is to build a relationship with them, so that that way when they do want the help, you are the person that they're going to call." Lack of clarity of coach role Many patients, when asked directly, indicated that they were uncertain about what the coach did and how the coach fit into their care team. For most, this was not concerning. For some, the lack of clarity limited their relationship with the coach. One homeless patient believed that "moral support" was all the coach had to offer, and he wanted help accessing housing.
Many coaches found that patients were asking them for help that they could not provide, including assistance at night and on weekends, or were trying to be friends with them. Some patients used substances near them or discussed drugs in a way that glorified use, which the coaches found challenging. Almost uniformly, the coaches said having physicians ask them to see patients with severe psychiatric illnesses or struggling with active suicidality or "What happens is the medical degrees kind of overshadow practical experience."-Coach [17] "[The medical team] don't know what to do. They looked at me like, 'Well, that's why we hired you. Like, figure it out.'"-Coach [17] psychosis was outside the scope of their training and was one of the hardest aspects of their job. Coaches found that some physicians had unrealistic expectations about their role, anticipating that the coach would be a "magic sobriety wand." Tension between coach and care team Patients were almost uniformly positive in their discussion of how coaches and physicians related, each playing a unique role in their recovery. Some patients indicated that they were glad that their coach did not share everything with the care team, as they wanted personal details kept confidential. Similarly, coaches said they tried not to share patient information with the team unless the patient was in danger or there were concerns about diversion. Nearly all coaches indicated that one of the greatest challenges of their job was working with physicians and within the medical setting. They often felt that physicians did not welcome their input about aspects of care planning. They also worried that the physicians, who relied on "very specific research papers" and were "set in their ways," did not take their lived experience seriously. More broadly, many coaches said they did not feel comfortable in a hospital or office setting. A number of coaches discussed dressing down or requesting permission from the clinic to wear jeans to emphasize similarities between themselves and their patients.

Discussion
This study explored patient and RC perspectives on the role of a RC in the management of a range of SUDs in an urban, primary care setting. Our findings allow us to define the activities and strengths of RCs. They also reveal 3 broad themes: (1) RCs filled gaps in primary care, offering a type of contact that is not otherwise available; (2) the therapeutic relationship between coach and patient was central to the patient's response to the coach; and (3) ambiguity about the RC role contributed to perceptions of tensions within the care team. Few prior studies have investigated the RC role, and ours is the first to examine the integration of RCs into primary care and include the views of RCs and patients-individuals directly affected by SUD interventions, but often marginalized in the discourse about them.

RCs fill gaps in primary care
Many patients and coaches perceived a power imbalance between physicians and patients that made patients feel uncomfortable opening up about their health, a dynamic that has been previously observed 31 and may be linked to worse health outcomes. 32 Both patients and RCs noticed that the coach reduced the power differential and marginalization patients felt within the medical system. Additionally, by connecting patients with social services, coaches helped address social determinants of health, which can affect health outcomes for people who use drugs [33][34][35] and may otherwise go unaddressed.
Finally, and perhaps most importantly, the coaches provided comprehensive social support. The 4 coach activities span Veiel's 4 theoretical dimensions of social support, from emotional support (relationship building) to more tangible instrumental support (system navigation) and from "everyday" support to more problem-oriented "crisis" support (harm reduction, supporting behavior change). 36

Therapeutic relationship
Most respondents believed that the relationship between patient and coach was central to the success of the intervention. Even if they were unsure exactly what the coach's role was, patients were willing to reach out and follow RC advice because they felt comfortable with the coach. This comfort emerged not only from shared experiences, but also from coach empathy, appearance, speech patterns, and accessibility. Our findings correspond with literature indicating that the therapeutic alliance has a modest effect on clinical outcomes in substance use care. 37 We hypothesize that the therapeutic relationship may be a more powerful predictor of efficacy in interventions with peer coaches, whose expertise is based on lived experience, rather than formal training. Notably, drawing from Veiel's framework of social support, patients often perceived RCs as "natural" support providers, like a family member or friend, whose role is not solely to provide unidirectional support, but also to receive support and be present at times when no support is needed. Conversely, the coaches perceived themselves as "institutional" providers of support, whose professional role is to give support, such as therapists or social workers. 36 Patients and coaches consistently emphasized that the shared experience of SUD was important for initial buy-in and may not be present with other sources of support.
Lack of agreement between physicians, coaches, and patients about the RC role, a tension that was often invisible to patients, placed considerable burden on coaches. SUD treatment often has high rates of staff turnover. 37 Although data on the effect of staff turnover on patient outcomes are equivocal, 37 high staff turnover can shape overall program costs, staff morale, and cohesion. 38 Our findings indicate that attention must be paid to creating a supportive work climate to retain RCs. Table 4 provides suggestions sample communication expectations for providers and RCs.

Limitations
The findings of this study must be interpreted in light of a number of limitations. First, this study was conducted within a single hospital network and may not be representative of settings not connected to an academic medical center. However, we included 3 satellite community health centers and a homeless health clinic, locations that are less well-resourced and typically see patients of low socioeconomic status. Second, the RC training and job description are specific to our state and institution, respectively. These may have shaped the RC perspectives and role and may limit the application of these findings to distinct settings. However, we believe that these findings could be used to guide RC training and job descriptions in other settings. Third, patients and coaches may not have been forthcoming about the challenges of the RC program in an interview, limiting the validity of our data. We were, however, able to capture a wide range of opinions of the program, suggesting minimal social desirability bias. Fourth, as primary care providers typically underdiagnose SUDs, 24 this study likely included primarily patients with more severe SUDs, those who would be most easily identified in primary care.

Directions for future research
Quantitative studies with a larger sample should be used to further define the RC role and examine the effects of RCs on outcomes including hospital readmission, treatment retention, and adherence to pharmacotherapy. Future studies should examine what training and lived experiences make RCs most effective. Broadening the pool of possible coaches may facilitate scale-up of RC programs. Finally, in order to fully understand the integration of RCs into primary care teams, qualitative studies should explore the perspectives of physicians and other team members.

Practice implications
These findings have a number of implications for practice. First, future recovery coaching programs should set clear expectations about the coach role with patients, coaches, and physicians upon initiation of the program and iteratively modify these expectations based on feedback from all parties. Second, RC training programs could benefit from an increased focus on psychiatric emergencies, referral to psychiatric treatment, appropriate outreach to social services, and self-care for coaches. Third, there has been considerable policy focus on integration of RCs into emergency departments. 7,39 This role would fundamentally differ from an RC in primary care, where the longitudinal relationship, over numerous contacts in a community setting, is central to coach work and perceived efficacy.

Conclusions
Our findings indicate that many patients see RCs as important members of the health care workforce, and RCs envision their roles and numbers expanding to meet the need of the country's SUD epidemic. The RC role is to act as a peer supporter, providing advice, motivating behavior change, and helping patients navigate the social, legal, and medical systems that impact their recovery. As these peer roles expand, health care teams will need to adapt to more seamlessly incorporate RCs. Although there is great need for more research into the effect of RCs on health outcomes, the role has potential to improve access and outcomes for patients struggling with substance use. Invite the coach to attend an appointment to meet the patient and demonstrate the importance of the coach's role encouraging coaches to take the lead in introducing their role.
In collaboration with the rest of the care team, develop a written description of the coach role that the provider can share with patients. Consider reinforcing boundary setting (times the coach is available, types of services the coach provides) by coaches during the initial visit.
Many patients are reluctant to ask for help because they don't know what services coaches offer or they assume they must be interested in recovery to work with a coach. Provide a clear description to patients about the range of services and supports you can provide.
If the patient will meet the coach at a separate time, ask the coach how their role should be described to the patient before this meeting.

Confidentiality
A provider statement to the patient supporting patient-coach confidentiality is a strong signal of respect to both patient and coach.
Work with your providers to develop a list of "red flag" information that should be communicated to providers.
Work with your institution to establish a peer mentor program and clear supervision structure for coaches that offers support and guidance around difficult decisions about disclosure.
Discuss difficult issues around disclosure with a peer mentor and your supervisor.

Shared decisionmaking
Agenda setting before or early in a shared appointment should include both patient and coach priorities.
Work with patients to develop a list of topics to discuss with a provider during an upcoming appointment. During shared appointments, use teach-back with patients and coaches to get an assessment of the collective understanding of a treatment plan.
During appointments, encourage patients to ask questions of providers when there may be misunderstanding from either the provider or patient. Involve coaches in multidisciplinary discussions of the management of complex patients in recovery.
If conflict is developing with a provider, work with a peer mentor or supervisor to develop ways of refocusing discussions on patient care. Integration with care team Invite coaches to develop brief teaching sessions for providers about addiction and recovery.
Ask a supervisor how you should document your interactions with patients in the her. Develop clear guidelines for the coach's notes in the electronic health record.
Engage with provider-led community-based research efforts.
Elicit feedback from coaches as a way of reflecting on behaviors or language that may be perceived as judgmental.
Ask for feedback from providers and peers.
Develop clear protocols for how RCs should handle psychiatric emergencies.
interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.